Acoustic seafloor sediment classification using self-organizing feature maps

Abstract

A self-organizing feature map (SOFM), a kind of artificial neural network (ANN) architecture, is used in this work for continental shelf seafloor sediment classification. Echo data are acquired using an echosounding system from three types of seafloor sediment areas. Excellent classification (approx. 100%) for an ideal output neuron grid size of 15 x 1 is obtained for a moving average of 35 input snapshots